Rolling out the barrel

28 July 2004

A drum containing Magnox fuel cladding was found to have been sealed for many years, allowing a build-up of potentially explosive gases which had to be safely released. By Phil Rhodes and Barry Vernon

During the clearance of building B280 at Sellafield, it became necessary to remove a Magnox Encapsulation Plant 500 litre stainless steel drum that had been stored for a period of approximately five years. The drum, manufactured from 3mm thick stainless steel, contained Magnox cladding and uranium. Typically, the contents of these drums would be submerged in water; a rubber gasket and a lid with five small ventilation holes would then be fitted.

Historically it was shown that the drum had been shipped from Winfrith to Sellafield, for the duration of the journey the vent holes in the lid were fitted with transit plugs to prevent spillage. After arriving at Sellafield, the plugs were to be removed, thereby purging any small quantities of gas which had been produced during transit. Unfortunately, once stored in building B280, the plugs were not removed and over the next five years a considerable build-up of gas was anticipated in the ullage of the drum which was calculated at 60 litres by X-ray inspection. Also, the drum had been stored on a plastic pallet adjacent to a chemical lab. The plastic pallet was deforming under the weight of the drum, and had developed an incline of about 5˚ (see Figure below).

Hydrogen is produced as a result of the corrosion of Magnox swarf in water; furthermore, it was believed that some of the drum’s contents were exposed above the surface of the water. Should any of these exposed items contain uranium fuel, then it would be possible to produce uranium hydride in an already hydrogen-rich atmosphere. Uranium hydride is pyrophoric and so will combust in the presence of oxygen, therefore, the drum potentially had a hydrogen-rich atmosphere and an ignition source in its ullage.

As hydrogen requires only a very low energy source to ignite, it was not possible to simply remove the now corroded vent plugs due to the high risk of a local spark. Furthermore the ingress of oxygen would increase the risk of spontaneous combustion of any uranium hydride. The drum was therefore to be purged while submerged in water; the release of gas pressure was to be followed by the subsequent flooding of the drum with water to purge any remaining gas. Only at this point could the vent plugs and the lid be removed safely and the drum could be transported for process of its contents.


Due to the low energy ignition characteristics of hydrogen, mechanical methods of drum piercing were discounted. This included any abrasive cutting media. Hazop studies concluded the drum was to be vented using a high-pressure water jetting system. Harsh Environment Systems’ (HES’s) remit was to design, manufacture, test, install and carry out the venting and drum opening operations.

The deviation from vertical in the drum’s orientation was only in one plane. Hazop studies concluded the drum must not be moved or subjected to any impact during gas release operations due to the volatile nature of the contents. Prior to operations, the drum had to be stabilised and prevented from moving or toppling. Any equipment used within the area of drum venting operations was required to comply with EC directive 94/9/EC ATEX (Atmospheres Explosibles).

Temperature gradients local to the water cutting site needed to be evaluated and, as the cutting water pressure was 250MPa, the internal pressure created within the drum at breakthrough was a concern. No data was available from the system suppliers and so the data was gathered on a representative test rig.

Suitable apparatus was also required to maintain the operational water level above the level of the drum in the event of seal failure or mains water pressure failure.

A robust, non-intrusive method of ensuring all gas contained within the drum had been purged, had to be developed prior to removing the lid.

Explosion risks during operations would never be zero so blast protection systems providing personnel/building protection and debris screens to contain the spread of radioactive debris in the event of an explosion were incorporated into the design.

Project risk and operational time was to be minimised by the use of modular design – all on-site systems had make and break connector based interfaces. Systems proven on the test rig would be the systems used on site.


High-pressure water jet cutting is a reliable and robust method of cutting stainless steel, usually coupled with Garnet grit injection. A discussion with the water jetting systems manufacturers, however, confirmed the grit was not necessary due to the thin section of the drum.

As the drum’s position on the pallet was unstable, localised stabilisation was necessary. It was decided to encapsulate the drum and pallet with grout. This also sealed the dedicated bund/blast protection. Curing of the grout caused localised heating because of the exothermic reaction as the grout solidified, local drum temperature gradients were monitored at the trials stage to select the correct grout mix characteristics.

As the system had to be portable, the design of the system-integrated components was perceived as a vital element in the success of the project. All components had to be modular, with rapid connections at interfaces. All portable systems had to be free of the need for specialist assembly or tooling. Operational constraints required the system to be installed and commissioned in less than 16 hours – a further driver for a modular, portable design.

The following design features were included in the design:

• The outer bund was designed to give maximum clearance between the drum and bund wall. It was also designed to use the adjacent chemical lab wall as a fitting guide.

• The outer bund was 1.75m tall, this allowed the drum to be totally submerged to a minimum depth of 200mm.

• After fitting the bund, the lower 350mm was filled with grout, this encapsulated the pallet and lower drum assembly, thus giving stability as well as providing a watertight seal between bund and drum.

• Preliminary trials were carried out to determine what pressures and flow rates were required to pierce the drum without the abrasive garnet material. These were found to be 27 litres of water per minute at a pressure of 250MPa.

• All camera equipment had to have low power integral lighting and be waterproof and intrinsically safe.

• Venting operations were to be carried out in the early hours of the morning, fire and police services were to evacuate and control the limited entry of personnel into controlled areas during venting operations.

• All equipment local to the venting operation should be hardwired and intrinsically safe with no junction boxes or power sources within 30m of the venting area. The control and monitoring station was situated 50m away.

• A 500 litre header tank would maintain a constant water level in the event of water mains or bund seal failure.

The exact moment of drum perforation was determined by :

• Video. Underwater video showed much agitation and little picture clarity during cutting. Upon breakthrough the agitation ceased and a clear pictured emerged showing the drum with the pressure jet clearly going through the side wall.

• Sound. An underwater microphone was attached to the drum. A harsh, high pitch noise was apparent during cutting. As the hole progressed through the material, the pitch became even higher – culminating in an sudden shrill as the jet broke through.

• Time. This was perhaps the most significant parameter; accurate to within approximately three seconds, one could anticipate the exact moment at which the jet would break through. This parameter combined with any one of the previous two, made the cutting operation very predictable.

The relevant data acquisition equipment was set up on a full-scale mock-up and all trials carried out were relevant to equipment and conditions that were to be encountered during the actual operation.

Two ultrasonic sensors were mounted on the drum, these were submersible and required calibration in situ in order to achieve a data point from which to operate. The first sensor was mounted on the drum side wall close to the underside of the lid, whilst the second was mounted on the lid at the highest point produced by the 5˚ incline.

A large explosion debris shield was designed and manufactured. In the event of an explosion, water and debris from the drum would travel upwards, hitting the underside of the shield assembly, which would contain and localise any contamination. The shield also provided a mobile platform for the water header tank and lighting for assembly operations prior to venting.

The vent holes were to be positioned as close to the underside of the drum lid lip as possible. These holes were pre-marked and an indicator pointer attached to the jetting nozzle clearly showed the hole to be cut. These cutting positions were also replicated on a XY grid displayed to controllers. In the event of video failure, the XY grid would display the nozzle position and the use of time and sound would determine at what point the drum had been perforated. These safeguards were proven successfully without the use of video.


A comprehensive series of trials and tests were performed at HES facilities on a fully representative (without the active components) model of the site area.

Temperature sensors were fixed to the inside surface of the drum to detect any temperature gradient, the sensors were logged via a data acquisition system but indicated only a very low gradient (less than 3˚C). Internal ambient pressure of the drum was also recorded, at breakthrough the pressure increase inside the drum was very low (less than 0.025MPa).

Trials provided the operations team the opportunity to rehearse and fully familiarise themselves with the systems and equipment. A final series of trials and training exercises were performed just prior to site operations to ensure all team members from both HES and BNFL were aware of the operating procedures and each individual’s responsibilities.


Once trials were complete, the following changes to the operation method were introduced:

• The distance of the cutting nozzle stand-off was altered after a series of trials. Initial figures suggested a stand of 3-5mm, however a stand-off of 25-30mm produced a cleaner and larger hole; this aided gas release and water ingress.

• During cutting operations it was found that the water agitation was so ferocious that air bubbles were being induced at surface level and travelling around underwater. This presented the problem of possible oxygen ingress into the drum upon breakthrough. The depth of the drum below the water surface was consequently increased to 300mm, a depth sufficient to relieve the problem.

• Once a complete system of operation had been produced and tested, a full trial was witnessed by BNFL and the Nuclear Installations Inspectorate. After no significant comments or concerns were raised by either party, the project was given the go ahead to proceed.


The final stage of the project – the installation of the equipment on site – was the culmination of 12 months work and investigation. The weeks leading up to the cutting operation saw the installation and grouting of the outer bund, the attachment of the cutting head and all its ancillaries, positioning of the blast shield and water level control systems requiring numerous equipment and safety checks. All this work was initiated and carried out by HES and qualified by BNFL.

On 23 October 2002, work began at 1730 with mobilisation to site of HES personnel and the trailer-mounted ISO freight container housing the intrinsically safe 220kW pump unit. The pump was positioned, as previously rehearsed, outside the controlled area. HES’s operations team then installed all temporary high pressure hoses and communications systems leading into the controlled area and into building B280 along with all other outstanding safety systems connected with the pumping unit. After the prerequisite safety and equipment checks, and with a significant section of the site evacuated, the venting operation was ready for commencement at 0100 on 24 October. With just four people within a 150m radius of the drum, the signal was given to cut the first vent hole.

After 1 minute 35 seconds, as was the case so many times before during the rehearsals, came the telltale signs that the first hole was about to break through. A short period then followed to allow the initial gas pressure release, the jetting head was re-indexed and the remaining holes were jetted. All that remained now was for the drum to fill with water and the level sensors to show that it was safe to release the lid.

As predicted, after approximately an hour and a half, the level sensors indicated that the drum was full of water and the gas had been purged. The final part of the operation was to enter B280 wearing the appropriate equipment and in complete darkness. With the aid of an intrinsically safe torch, the lid was removed and wedged open by means of a series of bespoke lifting bars and clamps. After retiring from the building and the elapse of a suitable gas dispersal period, power was restored to the building and all equipment dismantled and removed from site. The entire operation was completed successfully and without a single problem by 0400.

Repeated trials were essential to provide the design team with data to be fed back into the design loop to develop the final system design. Realistic simulation of the operations provided the operations team with an invaluable insight to the methodology and hazard reduction methods to be employed during operations.

The use of robust well proven technology and expertise from hazardous offshore oil and gas installations was the key element in the project success.

Success of the project can be primarily accredited to the excellent communication between all parties and the determination to manage and mitigate what were significant risks.

Author Info:

Based on a paper presented at the ‘Radioactive Waste Handling’ meeting organised by the Institution of Mechanical Engineers and held in Warrington, UK on 10 February 2004. Phil Rhodes and Barry Vernon, Harsh Environment Systems, Ennerdale Mill, Egremont, Cumbria CA22 2PN, UK

Related Articles
A new phase for a sound examination technique

Stages in venting system assembly

As each of the following milestones were met, constant checking and monitoring of the project ensured that every effort had been made in order to achieve a fully functional and fault free system on the night of operations.

• Mark five cutting targets onto the side of the drum.
• Accurately lower and position the outer bund over the drum and pallet.
• Grout drum, pallet and outer bund assemblies to building floor.
• Attach jetting head to purpose built transom on outer bund and align with cutting targets.
• Attach ultrasonic level sensors, microphones, cameras and lights.
• Position and lock debris blast shield into position.
• Attach water level control system from header tank on top of blast shield to outer bund assembly.
• Attach water exhaust duct, route into active drain and fill outer bund.
• Attach water main to header tank. Level in header tank maintained by ballcock.
• Assemble control station in adjacent building and lay umbilical to jetting system.
• Calibrate ultrasonic sensors.
• Install pump unit and attach all services and communications.

Privacy Policy
We have updated our privacy policy. In the latest update it explains what cookies are and how we use them on our site. To learn more about cookies and their benefits, please view our privacy policy. Please be aware that parts of this site will not function correctly if you disable cookies. By continuing to use this site, you consent to our use of cookies in accordance with our privacy policy unless you have disabled them.